The nature of hydrogen in x-ray photoelectron spectroscopy:
General patterns from hydroxides to hydrogen bonding
S. J. Kerber
Material Interface, Inc., Sussex, Wisconsin 53089-2244
J. J. Bruckner
University of Wisconsin–Milwaukee, Milwaukee, Wisconsin 53201
K. Wozniak
University of Warsaw, Warsaw, Poland
S. Seal, S. Hardcastle, and T. L. Barr
University of Wisconsin–Milwaukee, Milwaukee, Wisconsin 53201
~Received 11 October 1995; accepted 25 March 1996!
Important progressive alterations in chemical bonding are often realized through correlations with
shifts in the x-ray photoelectron spectroscopy ~XPS! binding energies of key elements. For example,
there are useful general XPS shifting schemes for such systems as oxides, nitrides, halides, and even
various functional groups in organics. Very general patterns, based upon location in the periodic
table, exist for many of these materials even when the structure is not strongly considered.
Unfortunately, apparently because of the lack of direct XPS detection of hydrogen, there seems to
be no general statements in the literature for describing hydrogen-containing compounds, despite the
fact that synergistic shifts obviously exist in the XPS spectra of elements attached to hydrogen ~e.g.,
for M–O–H vs M–O–M units, where M is a typical metal!. While not attempting a complete
review paper, in the present work we use XPS shifting patterns to evolve a series of interrelated
covalency/ionicity arguments to help explain the progressive, periodic changes in XPS peak
locations for such common cases as M–O–H- and M–N–H-containing systems. These arguments
are followed by consideration of the less dramatic XPS shifting patterns exhibited by metal and
metalloid hydrides, including organic bonding. The formalism concludes with a discussion of
hydrogen bonding detected by XPS. After a select review of the infrequent use made by others to
attribute XPS peak shifts to hydrogen bonding, we consider in some detail two cases recently
published by members of our group. One case involves the formation of –N–H–N– bonds in proton
sponge organic systems, while the other uses XPS to examine the formation of surface oriented
–O–H– – –O– bonds in the adsorption of peptides on oxidized metals. In the present article, the
XPS patterns for these two seemingly divergent cases are explained by closely related arguments
that may have far reaching generalities. © 1996 American Vacuum Society.
I. INTRODUCTION
To inorganic chemists, hydrogen chemistry is recognized
as one of the most regular and progressive features of the
periodic table. Because of the abundance of hydrogen in our
planetary system, this chemistry is one of the most prevalent
in our experience. Further, hydrogen has a propensity to terminate the chemistry of many bonding environments, making it a natural feature of many surfaces. While x-ray photoelectron spectroscopy ~XPS! is recognized as a preeminent
tool for surface chemical analysis, a major shortcoming is
that it cannot see hydrogen directly. Although the indirect
XPS registration of the effects of hydrogen have been recognized for many years, the pros and cons of hydrogen analysis
with XPS have not been collected and analyzed. It is our
intent to use a number of specific examples to try to explain
the general patterns of the XPS binding energies found in
hydrogen chemistry.
The compounds containing hydrogen fall naturally into
three classes: ~a! the molecular hydrides, which include those
systems containing –OH and –NH groups, ~b! the salt1314
J. Vac. Sci. Technol. A 14(3), May/Jun 1996
like hydrides of the alkali and alkaline earth metals, and ~c!
the interstitial hydrides of the transition metals. To this classification, we will follow Wells1 and Pauling2 and add a
fourth group designated as that involving the hydrogen bond,
i.e., X–H–Y, where X and Y are commonly either nitrogen
or oxygen.
In this article we will begin with the molecular hydrides,
which include ternary hydrides such as M–O–H and
M–N–H. This discussion will be followed by a consideration of the metal and metalloid hydrides, including the vital
class of carbon–hydrogen. The description culminates with
what may be the most important case of all—hydrogen bonding, including our new evidence for the XPS detection of the
presence of select hydrogen bonding, plus arguments for
general patterns of hydrogen bonding and chemistry. In all of
these cases the discussion will include a general interest in
the bonding chemistry with particular concern on the XPS
studies that have been accomplished for these systems.
For reasons of ultrahigh vacuum incompatibility, we will
omit all cases in which the HM bond is extremely ionic,
0734-2101/96/14(3)/1314/7/$10.00
©1996 American Vacuum Society
1314
Kerber et al.: Nature of hydrogen in XPS
1315
1315
either H1 ~e.g., HCl! or H 2 ~e.g., NaH!. Similar restrictions
prevent the measurements or XPS detection of the strong
hydroxide bases ~e.g., NaOH! and the strong oxyacids ~e.g.,
H2SO4). However, certain features of these aggressively
bonded materials may still be discerned from extrapolations
of XPS data realized for more covalent systems.
II. MOLECULAR HYDRIDES
A. M–O–H and M–N–H bonds
A consideration of the chemical shifts realized by hydroxides can be arrived at by first examining the situation that
results in the formation of oxides. Ignoring for a moment the
influences of particular structures, one may consider a metal
oxide to result in the creation of three-dimensional lattice
structures containing M–O–M bonds. When this bonding
forms, it has been traditional to classify the M–O bond as
ionic. However, it should be apparent that even for the most
reactive of metals ~e.g., Rb!, the M–O bond that is realized
in oxide formation always has at least a small degree of
covalency.2,3 Thus it is more proper to consider the bonds
realized in oxide formation to be mixed ionic/covalent
bonds, even though they are generally dominated by the
former. The interjection of a different M unit ~M8, with a
different electronegativity! to create M–O–M8, polarizes the
bonding chemistry around the oxygen.4,5 As the M–O bond
becomes more covalent, the electrons become more diffuse,
effectively moving away from the oxygen. In XPS, it has
been shown that the position of the O 1s line in an oxide can
be a measure of the relative covalency/ionicity of the oxide
bond.3,5
In the specific case of M8 being hydrogen, i.e., the hydroxide cases, we will consider two features: first the
changes in bonding chemistry as registered by XPS in going
from M–O–M to M–O–H, and second, the XPS changes
realized for the hydroxides when the M is varied. These features have been previously considered by Barr.6 For the purposes of this discussion, M8 –O–M8 will be considered to be
H–O–H, or water. Little XPS work of solid water, i.e., ice,
exists, but studies achieved thus far suggest a very large O
1s binding energy of approximately 533–534 eV.3,7 This
large value ~relative to ionic oxides such as CaO with an O
1s of 529.9 eV! is consistent with the very covalent oxygen–
hydrogen bond in neutral water. Based upon the arguments
presented so far, one would anticipate that the interjection of
the mixed M–O–H system will shift the O 1s from the water
value of 534 down toward the values for the more ionic
systems. In general, this means that the O 1s binding energy
for hydroxide systems should occur between the O 1s value
for the metal oxide and that for water. Data for experimental
systems are included in Table I. Although it must be pointed
out that this argument suffers from being exclusively initial
states in form and ignoring final state shifts,3,7,8 and although
a few exceptions have been referenced in the literature,9 this
TABLE I. Experimentally obtained XPS binding energies ~in eV60.1 eV! of select hydrogen-containing compounds and relevant nonhydrogen analogs. The
data included are from variable sources which may have had dissimilar binding energy calibrations.
H-containing
compound
Element
Peak
O
O
O
Cu
Pd
Cd
In
B
Ba
Y
Zr
Nb
Ti
C
1s
1s
1s
2p 3/2
3d 5/2
3d 5/2
3d 5/2
1s
3d 5/2
3d 5/2
3d 5/2
3d 5/2
2 p 3/2
1s
O
1s
C
1s
B.E.
932.5
335.4
405.0
444.0
186.5
780.6
155.8
178.8
202.2
453.9
284.6
Non-hydrogen
analog
Compound
B.E.
Compound
B.E.
Ref.
Ni~OH! 2
Cu~OH! 2
Zr~OH! 4
Cu~OH!2
Pd~OH!4
Cd~OH!2
In~OH!2
B 10H14
BaH2
YH3
ZrH2
NbH2
TiH2
C–O–H,
alcohol
C–O–H,
alcohol
O
532.0
531.7
531.2
934.8
338.6
405.1
445.8
187.6
782.0
157.7
178.8
202.9
454.0
286.1
NiO
CuO
ZrO2
CuO
PdO
CdO
In2O2
530.0
530.3
529.9
933.7
336.9
404.2
444.9
C–O–C,
ether
C–O–C,
ether
O
285.7
6
6
6
6
6
7
7
17
7
7
7
16
17
13
532.5
13
289.0
13
287.7
13,19
399.5
135.2
7,13
3
532.8
289.3
i
i
—C* –OC
O
—C–OH
C
N
P
1s
1s
2p
JVST A - Vacuum, Surfaces, and Films
—C–N–H
130.2
NH 3
NaH2PO4
285.7
398.8
134.2
i
—C–NH
C–NH2
P 2O 5
1316
Kerber et al.: Nature of hydrogen in XPS
method represents a generally successful formalism for the
prediction of hydroxide bonding effects.
In forming the mixed oxide M–O–M8 system, the binding energies of both M and M8 are also generally altered. If
M is more ionic than M8, then M is found to become even
more ionic in the mixed system and the M cationlike unit has
its binding energy increased.3,5 Unfortunately, if M8 is hydrogen it cannot be directly detected, we can only look at its
effects on M and O. Since most M’s are most cationic toward
O than H, the formation of a hydroxide should be registered
in XPS by an increase in binding energy of both the O 1s
and the M peaks.3
This argument holds true to varying degrees for all metals. As anticipated, the similarity in the ionicity of the transition metals yields similar O 1s binding energies for the
different hydroxides ~see Table I!. For most transition metals
M, the main oxide peak occurs at approximately 530 eV and
the major hydroxide peak is shifted to a higher binding energy, usually by 1.2–2.3 V. As documented in the literature
by Barr3,6,10 and Brundle,11 an outer hydroxide layer is
present on the surface of native oxides of titanium, vanadium, copper, nickel, and others. One must be careful with
absolute identifications, however, because this hydroxide
peak can be coincidental with a lower intensity peak due to
chemisorbed, polarized O2, and at still higher binding energies, ~;534 eV!, an additional small peak is often present
due to chemisorbed water and weakly adsorbed oxygen molecules.
Continuing to the right-hand side of the periodic table,
one reaches the amphoteric oxides in groups III and IVA,
e.g., aluminum and silicon. In these cases, the ionicity of any
OH bonds that are formed is very similar to that exhibited by
the oxide bonds, e.g., ;50% for Si. Thus, for example, the O
1s and Al 2p for a-alumina are 531 and 73.9 eV, respectively, whereas the same binding energies for gibbsite
@ Al~OH!3# are 531.1 and 74.2 eV, respectively.3,7 Similar behavior is exhibited by silica and its corresponding hydroxides. Such amphoteric behavior dissipates as one goes down
the groups IIIA and IVA columns, with the oxides and hydroxides becoming more ‘‘metallic’’ in behavior, similar to
those of the transition metals.3,12
For organics, similar considerations are possible for systems containing the C–O–H and C–O–C bonds. In a
C–O–C containing ~ether! system, the C 1s binding energy
is about 285.6 eV and O 1s is about 532.5 eV. In the case of
C–O–H alcohols, the carbon becomes somewhat more ionic
and the C 1s peak therefore increases to about 286.1 eV. The
O 1s also increases to approximately 532.8 eV ~to reflect the
covalency of the O–H bond!.13 Thus, carbon is a weak example of a metalloid. In organic acid systems, such as formic
acid, these arguments do not necessarily apply because there
is a CvO group that affects the covalent/ionic character of
the balance of the –OH system in a more complicated fashion than the simple M–O–M8 arguments used above.14
As one goes further to the right-hand side and up in the
periodic table, the bonding between M and O in a Mx Oy
system becomes increasingly covalent. Eventually the large
J. Vac. Sci. Technol. A, Vol. 14, No. 3, May/Jun 1996
1316
O 1s value rivals that of water and finally exceeds it. Therefore, the argument given above for metals is reversed. Unfortunately there are not significant amounts of data in the
literature for these materials because of their ultrahigh
vacuum incompatibility, e.g., alkali and alkaline earth hydroxides. However, data for phosphorus has been generated
and organized previously by Barr3 to demonstrate these arguments as included in Table I.
The M–N–H situation tends to be similar to, but less
dramatic than, these M–O–H cases. Metal nitrides place nitrogen in an effectively 23 oxidation state, thus producing
relatively low N 1s binding energies. The N 1s for CrN
occurs at 396.8 eV and even the metalloid Si3N4 pushes the
N 1s down to 397.7 eV.7 It is anticipated that the N 1s
binding energy for the corresponding metal amines will lie
somewhere between that for the corresponding nitride and
ammonia ~398.8 eV!.7,15 The presence of the relatively covalent hydrogen/nitrogen bond in metal amines is predicted to
drive the metal binding energy upwards. As one moves to the
right-hand side in the periodic table the effects for nitrogen
mirror those for oxygen but are less dramatic. Examples are
presented in Table I for C–N systems. Very little information
is available on Px Hy systems, but one would expect that the
trends would follow those for the NH system.
B. Metal hydrides
A number of previous publications report detectable
chemical shifts for various metal hydrides, particularly the
commercially important intermetallic alloys.16 Much of these
data are controversial due to the fact that some systems adsorb hydrogen instead of, or in addition to, chemically binding with it. In addition, the commonplace interaction between these metals and O2 and/or H2O substantially
interferes with the hydride results. Further, many of the anhydrous hydrides of interest are corrosive to the steel of
vacuum vessels and copper gaskets.
Table I demonstrates that XPS can generally distinguish
between many elemental species and their corresponding hydrides, with the core peaks of the hydride generally exhibiting an increasing binding energy.7,17 This is obviously due to
the partial extraction of one electron from the metal by the
hydrogen to create the hydride. With regard to the two sides
of the periodic table, the shifts for hydrides on the left-hand
side ~metal! of the table often appear to be smaller than the
shifts for the hydrides on the right-hand side ~nonmetal! of
the table. This may be due to the higher electronegativity of
the nonmetallics. A word of caution is in order, however,
since many of the reported XPS studies of nonmetal hydrides
were conducted on gas phase samples rather than solids.
Examples7 include SiH4 with a reported silicon binding energy of 107.1 eV relative to an elemental value of 99.7 eV,
PH3 with a phosphorus binding energy of 137.3 eV relative
to elemental phosphorus at 130.2 eV, and H2S with a sulfur
value of 170.4 eV relative to elemental sulfur at 164.2 eV.17
Gas phase systems do not have work functions ~or Fermi
edges!, and if one were to assume values of about 5 eV for
the former the reported large chemical shift for these hydride
1317
Kerber et al.: Nature of hydrogen in XPS
materials essentially disappears. Establishing a proper Fermi
edge ~with appropriate treatment of charging! will reduce
these differences even further.
C. Carbon–hydrogen systems
The involvement of hydrogen with carbonaceous systems
has produced the largest number of chemical compounds in
existence, however there is a relative similarity of the chemistries for many of these compounds. Thus, it is not surprising that XPS finds only marginal differences between the
effects of C–H, C–C, and –CvC– bonds. As discussed earlier, the C–O–H bonding unit is relatively amphoteric and
for this reason, the differences in XPS binding energies in
going from M–O–M bonds to M–O–H that played a significant role in the previously described metal hydroxides are
not as substantial in the case of carbon-based systems. Despite these qualifiers, certain examples of XPS differentiation
have been found for carbonaceous systems with hydrogen.
For purposes of uniformity, the value of 284.6 eV is often
selected as a point of reference for the hydrocarbon part of
typical adventitious carbon, serving as a basis to describe all
Cx Hy systems. However, several groups18,19 have found that
not all Cx Hy systems exhibit a singular binding energy, with
values suggested to range from 285.3 eV ~complex aliphatics! to 284.4 eV ~graphite!. For air exposed systems, one
generally finds C 1s spectra with modest, but significant,
shoulders on their high binding energy side due to the presence of CO2 , various carbonyl-containing organics in our
environment, and substantial C–OH, a version of our previously considered M–OH units. The recently published book
High Resolution XPS of Organic Polymers by Beamson and
Briggs13 has extensive binding energy tables that provide
excellent registers of the peak positions and ranges for almost all of the common carbonaceous bond situations that
are realizable, including those experiencing the effects of
hydrogen. Using these tables, and our own data, we have
found repetitive, generalized binding energies for Cx Hy and
C–Z–H bonded systems where Z is typically O and N, as
listed in Table I. The principal Cx Hy peak is assumed at
284.6 eV. Several types of C–O bonded systems can be identified, and as one can see in Table I, we suggest that with
reasonable XPS resolution one can differentiate between
C–O–C and C–O–H bonds. In addition, we find that there
are noticeable binding energy positions where the C–N–H
~;285.6 eV! and OvC–N–H ~;287.7 eV! species are
found.
III. HYDROGEN BONDING
A. General discussion
Hydrogen bonding, –X–H–X8 –, is one of the least understood and most important concepts in chemical bonding.
It is the basic reason for the unusual physical properties of
water and for the three-dimensional double helical structure
of DNA strands. The hydrogen bond forces certain structures
to have particular crystalline behavior where often one might
not expect crystallinity.1,2 For example, the hydrogen bond
JVST A - Vacuum, Surfaces, and Films
1317
constrains protein molecules to their native configurations. In
the X–H–X8 model, X and X8 may be any species that is
significantly electronegative, the most common cases being
fluorine, oxygen, and nitrogen. Cases of mixed oxygen–
hydrogen–nitrogen bonds are also common in biological systems.
Hydrogen forms a covalent bond with only one species,
e.g., hydrocarbons or hydroxides.1,2 Thus hydrogen bonds, in
which hydrogen is bonded to two species, results in bonds
that are neither totally covalent nor ionic. Pauling recognized
that the hydrogen bond is primarily ionic in character.2 Hydrogen bonds may be weak, with a bond energy perhaps as
low as 2 kcal/mol. But recent studies have shown that particular forms, such as those discussed below, may have energies as large as 15 kcal/mol.20 Because of the relatively
small bond energy of the hydrogen bond and the small activation energy involved in its formation, the hydrogen bond
often plays a part in near room temperature reactions. In the
present circumstance, we will be considering two types of
hydrogen bonds: ~1! the bond formed between two nitrogen
atoms in complex napthalenic organic crystals, and ~2! the
bond formed between two oxygen atoms during the absorption of peptides onto hydrated metal oxide surfaces.
B. Hydrogen bonding—DMAN
1,8 bis~dimethylamino!naphthalene ~DMAN, Fig. 1! is
the parent molecule of a class of compounds known as proton sponges.21 The significant steric strain in DMAN is associated with the amine groups which are in close proximity.
The strain may be relieved by sequestration of protons from
mineral or organic acids, which lead to the formation of very
stable ionic complexes containing hydrogen bonded
N–H–N1 bridges. As a consequence, proton sponges have
enormous affinities for H1 and thus strong basicities, typically pKa512. Proton sponges are the analogs of certain
superacids. Although some nuclear magnetic resonance
~NMR! and other studies of hydrogen-bonded DMAN complexes have been published,21 little is known about the parent
molecule, especially in the solid state.20 Recently, some
members of our group have performed 13C and 1H magic
angle spinning NMR spectroscopy21,22 and XPS of
DMAN22,23 and its complexes.
A series of acid halide-substituted DMAN compounds
were prepared by adding discrete amounts of various acid
halides to DMAN powder in the liquid phase and then crystallizing the resultant complex to yield a structure that contains a hydrogen bond between the amino groups. The acid
halides employed were HCl, HBr, and Hl; all of these are
classed as strong acids. In this study, XPS was used to determine if there were detectable differences that could be
correlated to progressive features in hydrogen bonding.22,23
For a substantial period of time, an XPS spectrum for
DMAN itself was not obtained because it apparently sublimes at room temperature at approximately 1023 Torr.
Elaborate procedures were recently developed to circumvent
this problem and implemented to permit detailed XPS analysis.
1318
Kerber et al.: Nature of hydrogen in XPS
1318
tive H 1s shifts with H-bond formation. Features that are
unique to the non-HX induced results include informative
XPS peak patterns for the various anions. These anions in2
2
clude SCN2, BF2
4 , CCl3COO , ClO4 , and others. The results from those studies will be published at a later date.
The nitrogen from all of these types of hydrogen-shifted
structures are expected to be shifted to a higher binding energy relative to the original nitrogen because there is the
creation of a positive center, producing shifts similar to those
detected between ammonia and ammonium ions ~398.8 and
401.7 eV, respectively!. Thus the upward shift reflects the
formation of a quasication; the variances in that shift and
peak size based upon acid type apparently reflect the variation in the strength of the hydrogen bond. Two of the N 1s
peaks are indicative of the asymmetry previously ascribed to
the proton position in the diamine H-bonding linkage.21–23
Other features, such as the retention of unaltered DMAN and
the prospect of variable proton positions are possible explanations for the multiplicity of these surface oriented XPS
peaks. Further arguments are presented below on this subject.
C. Hydrogen bonding—peptide
FIG. 1. ~a! 1,8 bis~dimethylamino!naphthalene, DMAN and ~b! XPS high
resolution N 1s spectra of DMAN and its acid halide derivatives: I. HI, II.
HBr, III. HCl, and IV. DMAN. The shift from DMAN to the left-hand side
is speculated to be due to hydrogen bonding.
The carbon spectra ~not illustrated! for DMAN are somewhat broadened and centered at about 285.0 eV suggesting a
manifold of at least two peaks; one at ;284.6 eV ~the hydrocarbon C 1s! and one at a higher energy, apparently indicative of the C–N bonds. Unreacted DMAN exhibits a
narrow N 1s line at ;399.5 eV ~no hydrogen bonding!. The
nitrogen spectra obtained for the acid halide derivatives,
DMANH1X2 are illustrated in Fig. 1. The primary DMAN
nitrogen at 399.5 eV is now spread into a clear multiplet,
with the new peaks occurring 1–3 eV higher than the major
peak. At times only a single secondary peak seems to occur
whereas other cases exhibit an apparent manifold of peaks,
perhaps as many as four.22,23 Based on supporting XRD22,23
and NMR22,23 data and the direction of these XPS shifts we
argue that the nature of the occurrence of these secondary
peaks suggests their direct association with hydrogen bonding as a result of the acid substitution. The peaks seem to
shift in position and relative size with changes in the halide
series.
In addition to the hydrogen halides, hydrogen bonding in
DMAN has been achieved with a number of other acids with
similar results.22,23 XPS of these systems provides spectra
which in part repeat and vindicate the arguments used for the
hydrogen halide ~HX! substituted DMAN systems, i.e., posiJ. Vac. Sci. Technol. A, Vol. 14, No. 3, May/Jun 1996
A study regarding the adsorption of three peptides on two
titanium alloys has been previously reported by Kerber.24
The nature of the adhesion was found to be consistent with
the formation of critical hydrogen bonds. The peptides were
arginine–glycine–aspartic acid–serine ~RGDS!, arginine–
glycine–aspartic acid–alanine ~RGDA!, and arginine–
phenylalanine–aspartic acid–serine ~RFDS!. The titanium
alloys were cp-Ti and Ti– 6Al– 4V. After an exposure to
aqueous peptide solution for 26 h at 25 °C, the titanium
samples were rinsed, dried, and analyzed with XPS. Adsorption isotherms were obtained for the six systems by plotting
the adsorbed peptide N/Ti ratio over a concentration range of
0.0–0.2 mg/ml for the matrix of experiments. The shape of
the isotherms supported the hypothesis that the peptides were
bonded to the hydrated native titanium oxide surface via hydrogen bonding,24 presumably between surface hydroxides
and the peptide organic acid groups. Fourier transform infrared ~FTIR! analysis gave strong indications that hydrogen
bonding to the metallic surface had occurred.25
Typical titanium XPS spectra were obtained from both
alloy surfaces, with 2 p peaks occurring at 458.5 eV ~titanium oxide! and at 453.8 eV ~metallic titanium!.24 Because
the TiO2 signal was so strong, shoulder peaks from the expected surface oriented titanium hydroxides were found to be
obscured by the stronger oxide peaks. There was no significant change in the titanium spectra with peptide adsorption.
Peptide nitrogen was routinely detected at 400.9 eV; once
again, there was no significant change in the nitrogen spectra
with peptide adsorption.
The high resolution XPS O 1s spectra from RGDS powder, distilled-water exposed titanium, and titanium exposed
to 2 mg/ml RGDS are shown in Fig. 2.24 The primary oxygen
peak for RGDS powder occurs at 531.2 eV, coinciding with
CvO; a high energy tailing is also found, due apparently to
1319
Kerber et al.: Nature of hydrogen in XPS
FIG. 2. XPS high resolution O 1s spectra from ~a! RGDS powder, ~b!
Ti– 6Al– 4V exposed to distilled water, and ~c! 2 mg/ml RGDS adsorbed
from aqueous solution on Ti– 6Al– 4V. Region I corresponds to TiO2 and
CvO, region II to Ti~OH!4, hydrated oxide, and C–OH, and region III to
new features suggested to be due to Ti–O–H– –O– C units between the
hydrated titania surface and C–O bonds in RGDS.
C–OH oxygen at 532.5 eV. The O 1s spectra for the waterexposed titanium is primarily typical of the oxide portion of
TiO2 ~530.7 eV! with high energy tailing, reflecting terminal
hydroxide and aquation in a manner consistent with all air
exposed oxides. After adsorption, in addition to undisturbed
TiO2, a series of peaks ranging from 531.5 to 533.5 eV increased ~relative to the main oxygen peak at 530.7 eV! with
increasing peptide solution concentration. The balance of the
higher energy O 1s structure found in Fig. 2~c! was presumed to be due to the presence of, and chemical interaction
between the peptide and the hydrated titania surface.24
In inspecting Fig. 2, it is apparent that some of the three
distinct binding energy regions of spectra ~c! can be associated with corresponding oxygen-containing parts of ~a! and
~b!. Therefore, region I corresponds to the positions of
TiO2 and CvO, while the right-hand side of region II is
close to Ti~OH!4 and the left-hand side of region II is indicative of any hydrated metal oxide and C–OH. Region III,
however, does not reflect any portion of ~a! or ~b!. In view of
the higher binding energy of region III and the lack of other
candidates, the new feature is suggested to be due to Ti–O–
H– – –O–C units that may result from hydrogen bonding between the titania surface and RGDS. These features are clarified and the mechanisms for the attachment are described in
Ref. 24.
D. XPS of hydrogen bonding—general remarks
We are not the first to employ XPS to help to describe
hydrogen bonding. For example, Incorvia and Contarini26
used XPS to investigate the adsorption of select, complex
~inhibitor! amines on the surface of passivated iron. The system proves to be quite similar to our amine on Ti results.
Thus, retention of the obvious metal–OH structure results
with indication of the formation of Fe–O–H– –N–R hydrogen bonds. The latter are shown to produce higher binding
JVST A - Vacuum, Surfaces, and Films
1319
energy shifts in both the N 1s and O 1s peak manifolds
and a corresponding ~and quite interesting! lack of shift in
the Fe ~note the behavior of Ti in our studies!. On the other
hand, Nilsson et al.27 employed XPS to follow the metal
adsorption destruction of N–H–N-type hydrogen bonds.
Similar XPS shifts to higher binding energy were detected by
Bigelow et al.28 in their XPS study of the production of intramolecular hydrogen bonding.
The general features of our XPS-hydrogen bonding studies are the introduction of H into a complex diamine system,
DMAN, and the retention of hydroxides on air oxidized Ti to
form –O–H–O bonds following adsorption of proteins. Both
cases exhibit the same XPS patterns featuring positive shifts
in the amino or hydrated units ~–N–H–N– and –O–H–O–!
and reduction in the positive shifts by the species attached to
this unit ~the carbons in one case and titanium and carbon in
the other!. In view of our previous discussion of the influence of the alterations in the bond covalency/ionicity on
binding energy shifts, we turn to the conclusions of Pauling2
and Wells,1 who point out that the introduction of ~–A–H–
A–! hydrogen bonding produces a bond that is mixed of
type, but primarily ionic. This means that the balance of
charge around the A unit requires that following hydrogen
bonding the M–A bond must exhibit enhanced covalency. As
a result one should expect the binding energies of the A unit
to increase and that for M to decrease. As started above, this
is detected in both of our H bonding cases. The multiplicity
of AwN peaks in the acid substituted DMAN case primarily
results from the asymmetry of the H1 position between the
two amine nitrogen. Thus, the C–N bond that is closest to
the resulting proton is forced to be extremely covalent and its
N 1s exhibits an increasing binding energy above 401 eV.
The other C–N bond undergoes only a marginal increase in
covalency, indicative of the N 1s at or below 400 eV.
IV. CONCLUSION
The present study concerns the influence of hydrogen in
XPS. We have shown that although we are not able to detect
hydrogen directly in this form of spectroscopy, one is usually
able to detect and ascribe the effects induced by the presence
of hydrogen into the XPS spectra of the elements that bond
directly to it and even, in many cases, the elements ~if any!
that are bonded to the elements attached to hydrogen.
We have demonstrated these features by starting with a
description of hydroxides and amines and showing how their
XPS spectra differ from their corresponding oxides and nitrides. This discussion was followed by a cursory investigation of metal hydrides and how their XPS spectra differ from
those of the elemental metals. The cases of organic Cx Hy ,
C–O–H, and C–N–H units were then considered as a
unique collective set. The bonding for these systems obviously pivot on the unique properties of carbon and both its
inherent amphoteric nature and unique propensity for covalent bonding. Both of these features play a central role in the
relevance of carbon-based systems in hydrogen bonding—
the final subject of our article.
1320
Kerber et al.: Nature of hydrogen in XPS
In our study of H bonding we describe two quite divergent cases: one in which the H-bonding results from the substitution of H1 ions into a basic system and the other in
which there is evidence of H bonding following adsorption
of peptides on a surface hydroxylated metal ~Ti! oxide. Despite these obvious differences we have been able to show
that the resulting XPS binding energy shifts induced by H
bonding follow the same pattern, i.e., for M–A–H–A–M8
the binding energies of the A units are increased while those
for M and M8 decrease. This, we explain, results from an
enhancement of the covalency of the M–A bond induced by
the creation of the relatively ionic A–H bonds. Evidence is
also provided suggesting a substantial generality for these
arguments.
A. F. Wells, Structural Inorganic Chemistry ~Clarendon, Oxford, 1945!,
Chap.7.
2
L. Pauling, The Nature of the Chemical Bond, 3rd ed. ~Cornell University
Press, Ithaca, NY, 1960!, Chaps. 3, 12.
3
T. L. Barr, Modern ESCA ~Chemical Rubber, Boca Raton, FL, 1994!.
4
J. A. Duffy, J. Phys. Chem. Glasses 30, 1 ~1989!.
5
T. L. Barr, J. Vac. Sci. Technol. A 9, 1793 ~1991!.
6
T. L. Barr, J. Phys. Chem. 82, 1801 ~1978!.
7
Practical Surface Analysis, 2nd ed., edited by D. Briggs and M. P. Seah
~Wiley, Chichester, 1990!.
8
P. S. Bagus, C. R. Brundle, G. Pacchioni, and F. Parmigiani, Surf. Sci.
Rep. 19, 265 ~1993!.
9
S. W. Gaarenstroom and N. Winograd, J. Chem. Phys. 67, 3500 ~1977!.
10
T. L. Barr, J. Vac. Sci. Technol. 14, 650 ~1977!.
11
C. R. Brundle, Faraday Discuss. Chem. Soc. 60, 159 ~1975!.
1
J. Vac. Sci. Technol. A, Vol. 14, No. 3, May/Jun 1996
1320
12
F. A. Cotton and G. Wilkinson, Advances in Inorganic Chemistry, 5th ed.
~Wiley, New York, 1988!.
13
G. Beamson and D. Briggs, High Resolution XPS of Organic Polymers
~Wiley, Chichester, 1992!.
14
T. L. Barr and M. P. Yin, J. Vac. Sci. Technol. A 10, 2788 ~1992!.
15
F. P. Larkins and A. Lubenfeld, J. Electron Spectrosc. Relat. Phenom. 15,
137 ~1979!.
16
K. Tanaka, M. Ushida, K. Sumiyama, and Y. Nakamura, J. Non-Cryst.
Solids 117-118, 429 ~1990!; K. Tanaka and H. Aoki, J. Nucl. Mater. 169,
299 ~1989!.
17
C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder, and G. E. Muilenberg, Handbook of X-Ray Photoelectron Spectroscopy ~Perkin–Elmer,
Eden Prairie, MN, 1979!.
18
G. Barth, R. Linder, and C. Bryson, Surf. Interface Anal. 11, 307 ~1988!.
19
T. L. Barr, S. Seal, S. Krezoski, and D. H. Petering, Surf. Interface Anal.
~in press!.
20
A. Warohel, A. Papazyan, and P. A. Kollman, Science 269, 102 ~1995!;
W. W. Cleland and M. M. Kreevoy, ibid. 269, 104 ~1995!; P. A. Frey, ibid.
269, 104 ~1995!.
21
K. Wozniak, H. He, J. Klinowski, and E. Grech, J. Phys. Chem. 99, 1403
~1995!.
22
K. Wozniak, H. He, J. Klinowski, W. Jones, and T. L. Barr, J. Phys.
Chem. 99, 14667 ~1995!.
23
K. Wozniak, H. He, J. Klinowski, W. Jones, T. L. Barr, and S. Hardcastle,
J. Phys. Chem. ~in press!.
24
S. J. Kerber, J. Vac. Sci. Technol. A 13, 2619 ~1995!.
25
S. J. Kerber, H. Mueller, and T. L. Barr ~unpublished!.
26
M. J. Incorvia and S. Contarini, J. Electrochem. Soc. 136, 2493 ~1989!.
27
J. O. Nilsson, C. Tornkvist, and B. Liedberg, Appl. Surf. Sci. 37, 306
~1989!.
28
R. W. Bigelow, K. Y. Law, P. H. K. Pan, and H. J. Freund, J. Electron
Spectrosc. Relat. Phenom. 46, 1 ~1988!.